Retroviral Expression in Embryonic Stem Cells and Hematopoietic Stem Cells

Tissue culture.

ES cells were cultured as described previously (21). To generate ES cell clones for injection into blastocysts, the ES cells were maintained on irradiated mouse embryonic fibroblasts (MEFs) with 500 U of leukemia inhibitory factor (LIF) per ml (22). For other experiments, the ES cells were cultured without MEFs in 1,000 U of LIF per ml. 293 cells were maintained in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum (FBS), penicillin, streptomycin, and glutamine. Abelson virus-transformed B cells were maintained in RPMI 1640 supplemented with 10% defined FBS (HyClone), penicillin, streptomycin, glutamine, and 50 μM β-mercaptoethanol. ES cells with retroviral integrants were in vitro differentiated as follows: the cells were passaged without LIF in the absence of MEFs on bacterial plastic petri dishes for 4 days, trypsinized, and cultivated with or without retinoic acid for 2 weeks (25).

Plasmids.

The retroviral vectors MfgGFP, pMXGFP, and MSCViresGFP have been described elsewhere (27, 33, 37). The MSCViresGFP vector was modified by introducing either the Cre recombinase or the human Bcl-2 gene upstream of the internal ribosome entry site (IRES)-GFP cassette as described elsewhere (11, 37). The replication-incompetent helper plasmid pCL-eco was used (24).

Retroviral infections.

To generate retroviral supernatants, 293 cells were transiently transfected by calcium phosphate-mediated coprecipitation with 5 μg of the replication-incompetent helper vector pCL-eco and 10 μg of the reporter retroviral vector as stated elsewhere (28). The cells were fed at 24 h postinfection, and the retroviral supernatant was used at 48 h. The cells continued to produce high-titer retroviruses for another 2 days, and that supernatant was used if needed for additional experiments. The supernatant was collected, brought to 4 μg of Polybrene per ml–10 mM HEPES, and filtered (0.45-μm-pore-size filter) for use.

ES cells for infection were washed and trypsinized. They were then plated at 10⁶ cells per well of a six-well dish and centrifuged. The ES cell medium was removed, and retroviral supernatant was added at 1 ml/10⁶ cells. Next, the plate was centrifuged for 45 min at 2,500 rpm at room temperature. The retroviral supernatants were removed; the cells were resuspended in ES cell medium and plated onto gelatinized dishes. ES cells used to generate mice were plated onto irradiated MEFs.

Bone marrow was infected as follows (36). C57BL/6 mice were obtained from the Jackson Laboratory (Bar Harbor, Maine). Bone marrow cells were harvested from the tibias and femurs of C57BL mice 5 days after they received an intraperitoneal injection of 5 mg of 5-fluorouracil (Sigma) in Dulbecco's phosphate-buffered saline (Gibco/BRL). These cells were then cultured for 4 days at 2 × 10⁶ cells/ml with recombinant mouse interleukin-3 (rmIL-3; 20 ng/ml), rmIL-6 (50 ng/ml), and (50 ng/ml; recombinant mouse stem cell factor; R&D Systems) in Dulbecco's modified Eagle medium containing 10% FBS. After 48 and 72 h, the bone marrow cells were spin infected with the retroviral supernatant generated as described above. Then the retroviral supernatant was removed and replaced with growth medium containing cytokines.

FACS (fluorescence-activated cell sorting) analysis and sorting.

Adherent cell lines were trypsinized, washed, and resuspended in complete medium to achieve a single-cell suspension at the time points indicated. Nonadherent cells were used directly for analysis. Organs were disrupted manually and passed through a 70-μm mesh to generate a single-cell suspension. The cells were analyzed for viability using scatter properties and the exclusion of propidium iodide. The level of GFP expression was monitored by fluorescence without compensation to detect cells with low levels of GFP expression. The ES cells were sorted into ES cell medium and plated immediately onto either gelatinized plates or MEFs for blastocyst injections. The survival of ES cells after sorting was approximately 50%, as measured by the number of colonies generated divided by the expected number of colonies.

5-AzadC treatments.

ES cells were treated with 0.15 μM 5-azadC (Sigma) at days 1 and 3 postplating. The cells were fed, allowed to recover, and then assayed 4 to 8 days later. The red blood cells in whole blood were lysed (5), and the remaining cells were stained with fluorescently labeled anti-H2-b, anti-H2-d, anti-B220, anti-TCRa (Pharmingen) at 1:200 as indicated. At day 0, splenocytes were treated with either anti-CD3 or anti-CD40 (Pharmingen); 0.15 μM 5-azadC was added at day 1, and the anti-CD3-treated cells were assayed at day 4. 5-AzadC was added again to the B-cell cultures with fresh anti-CD40 at day 4, and the cells were assayed at day 6.

Staurosporine-mediated cell death.

ES cells were infected with the stated retrovirus and treated with staurosporine at day 4 postinfection for 24 h with the indicated concentration of drug. The percentage of viable, GFP-positive cells was determined by flow cytometry (6). Data are presented as a percentage of GFP-positive cells before treatment. Results from one representative experiment of three performed are shown.

LacZ staining.

ES cells were infected with the stated retrovirus and sorted for GFP expression at day 3 postinfection. The ES cells were plated and cultured for an additional 5 days and stained for LacZ expression as described elsewhere (41).

Adoptive transfers.

Recipient mice (10) received a total of 1,200 rads of whole-body radiation in two doses (800 and 400 rads) 3 h apart and were then injected with 2 × 10⁶ to 5 × 10⁶ infected bone marrow cells. Irradiated mice were maintained on trimethaprim-sulfamethoxazole in sterile cages for 4 to 6 weeks to prevent opportunistic infections (34). Serial passages were performed by harvesting bone marrow from mice 6 to 8 weeks postreconstitution and transferring 2 × 10⁶ to 5 × 10⁶ cells into irradiated recipients. Mice were analyzed 8 to 12 weeks posttransfer to allow reconstitution of the T-cell compartment. These experiments were repeated multiple times with similar results.

Southern blot analysis.

The genomic DNA was isolated as described elsewhere (19). Ten micrograms of DNA was digested with the stated restriction enzyme overnight. The products were resolved on an agarose gel, transferred to a nylon membrane, and detected using a probe that spans the entire GFP coding sequence.

High-efficiency retroviral expression in ES cells.

Retroviral vectors based on the MSCV LTR were constructed with a multiple cloning site followed by an IRES driving expression of the gene for GFP as schematically diagrammed in Fig. 1A (MiG) (37). We generated high-titer retroviruses by transient transfection and infected ES cells with an adapted spin infection protocol. Using this protocol, we reproducibly achieved high-efficiency (>50%) infection of ES cells as measured by flow cytometry; uninfected control cells were negative for GFP expression (Fig. 1B). The intracellular concentration of GFP is directly proportional to the fluorescence intensity measured by flow cytometry.

Next, we compared expression of the MSCV-based retrovirus and Moloney virus-based retroviral vectors in ES cells. GFP expression was detectable with the MSCV LTR-containing MiG vector but not with the two Moloney virus-based viruses pMX (27) and Mfg (33) (Fig. 1B). This was not due to inefficient genomic integration of the provirus or to a lower titer. Southern blot analysis of genomic DNA demonstrated that all three proviruses were integrated in the ES cells (Fig. 1C). Also, when parallel B-cell cultures were infected with the retroviral supernatant used to infect ES cells, all of the retroviral vectors were expressed in B cells at comparable efficiencies (Fig. 1B).

GFP expression driven by the MSCV LTR in ES cells was substantially lower than in other differentiated cell lines tested (Fig. 1B and data not shown) (20). To determine whether this low level of expression was sufficient to drive functional expression of other gene products, we cloned the Cre recombinase upstream of the IRES-GFP cassette to generate MSCVCreiresGFP. We tested for Cre activity by infecting ES cells that contain a translational stop sequence flanked by loxP sites located between the Rosa 26 promoter and a lacZ reporter schematically diagrammed in Fig. 2A (34). If Cre is expressed at functional levels in these ES cells, the protein will catalyze recombination of the loxP sites, leading to loss of the stop sequences and the expression of LacZ. Indeed, we found that >99% of GFP-positive ES cells that were infected with the Cre-expressing retrovirus were also LacZ positive (Fig. 2B). Uninfected cells were both GFP negative and LacZ negative (data not shown). This indicates that virus-mediated gene transfer resulted in functional Cre expression.

Because Cre activity is required only transiently for LacZ expression, we tested a second gene product that must be stably expressed throughout the experiment. It has been demonstrated that Bcl-2 expression protects many cell types against staurosporine-mediated apoptosis (10). Therefore, we examined whether Bcl-2 could protect ES cells from apoptosis when expressed from the MSCV LTR. We cloned human Bcl-2 upstream of the IRES-GFP cassette to generate MSCV Bcl-2iresGFP. Wild-type ES cells were infected with either the Bcl-2-expressing retrovirus or the control virus lacking Bcl-2. Increasing concentrations of staurosporine were added to the cultures, and flow cytometry was used to assay for both viability and GFP expression. GFP-positive cells infected with the Bcl-2-containing virus were significantly protected from staurosporine-mediated cell death compared to the GFP-negative cells or GFP-positive cells infected with the control retrovirus (Fig. 2C). Therefore, the level of expression from the MSCV LTR is sufficient for stable functional gene expression in ES cells.

Short-term transcriptional silencing in ES cells is methylation independent.

It long has been hypothesized that retroviruses are transcriptionally silenced in embryonic cells by DNA methylation (12, 14, 21). Therefore, it was possible that DNA methylation of the MSCV LTR was responsible for the decreased level of expression in ES cells compared to other cell types (Fig. 1B). In addition, we sought to test whether DNA methylation of the Moloney virus-based vectors in the wild-type ES cells was the mechanism by which the Moloney virus-based LTRs were silenced (9, 13). To this end, we infected ES cells deficient for the maintenance DNA methyltransferase gene, Dnmt1, the loss of which results in genomewide hypomethylation (21, 22). Dnmt1^−/− ES cells are demethylated, and proviral sequences remain unmethylated. The Moloney virus-based retroviruses such as pMX remained silent even when introduced into Dnmt1^−/− ES cells, whereas MSCV expressed similar levels of GFP in both Dnmt1^+/+ and Dnmt1^−/− ES cells (Fig. 3A). Therefore, the initial block in transcription directed by Moloney virus LTRs in ES cells is independent of DNA methylation and is presumably due to the binding of trans-acting factors. In addition, the mean fluorescence intensities of GFP were comparable between the Dnmt1^+/+ and Dnmt1^−/− ES cells, indicating that the basal level of expression of the MSCV LTR is independent of DNA methylation.

DNA methylation constrains long-term retroviral expression.

MiG-infected GFP-expressing ES cells were continually passaged to test the effect of DNA methylation on long-term expression. Though GFP expression was high in both Dnmt1^−/− and Dnmt1^+/+ ES cells at 5 days postinfection, a substantial fraction of the infected wild-type ES were GFP negative at 26 days postinfection. This was apparent by both a loss in the percentage of GFP-positive cells as well as a decrease in the mean fluorescence intensity of the bulk population of wild-type ES cells and was observed in both bulk cultures and individual cloned lines containing single integrants (Fig. 3B and data not shown). The fraction of GFP-positive cells continues to decrease with additional passages, as shown in Fig. 3C. These results suggest that long-term expression was suppressed by DNA methylation. To directly test whether retroviral repression was due to de novo methylation of the newly integrated retroviruses, we treated the long-term cultures with 5-azadC, a drug that leads to hypomethylation of genomic DNA (16). If DNA methylation was preventing expression of the MSCV LTR, treatment with the drug should activate retroviral expression. Indeed, we found that 5-azadC treatment of ES cells that had lost expression of GFP through long-term passage reactivated the provirus (Fig. 3C). In contrast, Dnmt1^−/− ES cells infected with the retrovirus did not lose expression of GFP; thus, treatment with 5-azadC did not significantly affect retroviral expression (Fig. 3C). We also analyzed clonal lines containing single proviral integrants in which GFP expression was progressively silenced and found that treatment with 5-azadC resulted in the reactivation of gene expression in all cases (data not shown). This demonstrates that DNA methylation controls long-term but not short-term expression of retroviruses in ES cells.

Expression is maintained after in vitro differentiation.

Previously, in vitro differentiation of ES cells had been demonstrated to silence expression of retroviral sequences (12, 20). Thus, we tested whether GFP expression from the MiG retrovirus in ES cells was affected by in vitro differentiation. We cultured MiG-infected wild-type ES cells in the absence of embryonic feeder cells and LIF in suspension to generate embryoid bodies. Disaggregated embryoid bodies were replated either with or without retinoic acid. We found no change in GFP expression in MiG-infected bulk cultures or individual subclones containing one to several integrants upon in vitro differentiation with either method, as shown for one clonal line containing multiple integrants in Fig. 3D. GFP expression was unchanged in all in vitro-differentiated ES cell lines, regardless of whether the subclones contained only a single or multiple integrants. This indicates that the MSCV-based MiG retrovirus is not silenced by in vitro differentiation.

Generation of mice from GFP-expressing MiG-infected ES cells.

We next determined whether expression of the MSCV-based MiG vector was affected by in vivo differentiation of the infected ES cells. Cells from the chimeric animals were derived by injection of MiG-infected wild-type ES cells (derived from 129-Sv/Jae mice) into BALB/c blastocysts. MiG-infected Dnmt1^−/− ES cells cannot be used for injection into blastocysts, because Dnmt1^−/− ES cells die upon differention and therefore do not contribute significantly to adult mice (22). MiG-infected wild-type ES cells were sorted for GFP expression by flow cytometry prior to injection, and two GFP-expressing clones, R2 and R11, were isolated (Fig. 4B). Southern blot analysis demonstrated that R2 contained two integrants that comigrate on an agarose gel, and R11 contained three proviral integrants (Fig. 4A). High-contribution chimeras (>80% by coat color) were generated from the R2 and R11 ES cells, which transmitted the proviruses to their offspring (data not shown).

To test whether the chimeras expressed the integrated retroviruses, we isolated peripheral blood mononuclear cells (PBMCs) from both the R2- and R11-derived chimeras. To distinguish whether the PBMCs were derived from the ES cell donor or the host blastocyst, we stained the cells with antibodies that recognized specific major histocompatibility complex class I haplotypes (Pharmingen). The donor ES cells (129 derived) are H2-b, and the blastocysts (BALB/C derived) are H2-d (Fig. 4C and data not shown). In addition, we stained the PBMCs with a pan-T-cell (TCRa) (Fig. 4C) or pan-B-cell (B220) antibody (data not shown) to determine the ES cell contribution to these lineages. Using this strategy, we found that approximately 90% of the PBMCs from either the R2 or R11 chimera were ES cell derived as measured by H2-b staining (Fig. 4C, data not shown). However, the majority of the cells did not express GFP in either chimera (Fig. 4C and data not shown). On the order of 0.1% of the PBMCs that were ES cell derived were GFP positive, compared to less than 0.01% that were blastocyst derived (Fig. 4C). Similar results were also obtained with cells from the R11 chimera (data not shown). The results indicate that the MSCV LTR is repressed during in vivo differentiation to lymphocytes. Nevertheless, a small number of cells escaped silencing and expressed GFP. This transcriptional repression of the MiG provirus in the chimeras is in contrast to the GFP expression both in the donor ES and after in vitro differentiation (Fig. 4B, data not shown).

To determine if other somatic cells expressed the retroviral integrants, we analyzed the progeny of the chimeras. We isolated spleen, thymus, kidney, and liver cells from an animal carrying the two proviral integrants present in the R2 chimera and a littermate control containing no retroviral integrants. We analyzed these cells for GFP expression by FACS analysis and found no detectable expression of GFP in the splenocytes, thymocytes, renal cells, or hepatocyes (Fig. 5A and data not shown).

In vitro reactivation of retroviral expression.

One possible explanation for transcriptional repression during in vivo differentiation was de novo methylation of the integrated retroviral LTR during embryonic development. To test this hypothesis, we cultured splenocytes from a mouse containing the R2 proviruses and from a littermate control, by treating the cells with either anti-CD3 or anti-CD40 to activate and induce proliferation of the T cells or B cells, respectively (4). We then assayed for GFP expression by flow cytometry and found that proliferation of the splenocytes did not activate expression of the retrovirus (data not shown). Next, we added 5-azadC to the splenocyte cultures to induce demethylation of the retroviruses. Indeed, treatment with 5-azadC activated expression in approximately 2% of the T cells (anti-CD3) (Fig. 5B) and 2% of the B cells (anti-CD40) (data not shown). In addition, when in vivo-differentiated cells, which had been isolated from the kidney of a transgenic mouse and transformed with simian virus 40 large T antigen (15), were treated with 5-azadC, activation of the silent provirus was observed in a similar fraction of the cells (data not shown). The extent of reactivation of expression of the provirus in in vivo-differentiated cells by 5-azadC was lower than in ES cells, where the reactivation of the provirus with 5-azadC was almost complete.

We next determined whether demethylation of the retrovirus in vivo would activate expression of the integrated retroviruses (13). Newborn mice were subcutaneously injected with 5-azadC at postnatal day 5 and subsequently analyzed at postnatal day 14 for GFP expression by flow cytometry. We found that 5-azadC-injected animals but not the uninjected controls had activated GFP expression of the proviruses in the spleen, thymus, and kidney (Fig. 5C and data not shown). When we injected higher concentration of 5-azadC in an effort to further demethylate the newborn mice, all injected animals died. This result demonstrated that repression by DNA methylation is, at least in part, responsible for silencing expression of the retroviral LTR in vivo.

Retroviral expression in HS cells after serial adoptive transfers.

Bone marrow contains the HS cells that can stably repopulate the hematopoietic system after transfer to lethally irradiated mice. To determine whether HS cells can be effectively transduced and express the MiG retrovirus, we used infected bone marrow cells to reconstitute lethally irradiated mice (Fig. 6). We found that between 30 and 80% of the splenocytes from these primary recipients expressed the retrovirus, as measured by FACS analysis for GFP expression and shown for one representative experiment (Fig. 6A). The MiG virus was expressed in the B-cell, T-cell, and granulocyte compartments, as measured by a pan-B cell (B220), pan-T-cell (Thy-1), and pan-granulocyte (Gr-1) marker electronically gated on GFP-positive cells (Fig. 6B and data not shown). Because a large fraction of the splenocytes in the primary recipients are derived from relatively differentiated, lineage-committed progenitors, serial adoptive transfers are required to test for retroviral expression in the true HS cells (17). Therefore, we used bone marrow from these primary recipients to serially reconstitute lethally irradiated mice. This protocol requires substantial expansion from the stem cells and tests for long-term expression of the retrovirus. We observed no change in the percentage GFP-positive HS cells, and the level of GFP expression from the adoptive transfers into multiple recipients was stable over three additional passages (4° recipient). In addition, the infected cells gave rise to both B- and T-cell lineages at the expected ratios (Fig. 6B), demonstrating not only that the MiG retrovirus transduced the long-term repopulating HS cells but also that the MiG-mediated GFP expression was stable during in vivo hematopoietic differentiation. However, our results do not exclude the possibility that in addition to the transcriptionally active proviruses present within these cells, there are also copies of the virus that were transcriptionally silenced.